Research ArticleCancer

Avasopasem manganese synergizes with hypofractionated radiation to ablate tumors through the generation of hydrogen peroxide

See allHide authors and affiliations

Science Translational Medicine  12 May 2021:
Vol. 13, Issue 593, eabb3768
DOI: 10.1126/scitranslmed.abb3768

Reinforcing radiation therapy

The superoxide dismutase mimetic avasopasem manganese (AVA) is in a late-stage clinical trial for its ability to limit toxicity in the normal tissues of patients with cancer during radiation treatment. Now, Sishc et al. show that AVA also sensitizes cancer cells to high dose per fraction radiation by enhancing the generation of hydrogen peroxide, inflammation, and apoptosis. The combined modality had synergistic effects on xenograft tumors in mouse models but was not cytotoxic to nonneoplastic cells. This study indicates that AVA may have therapeutic, not just protective, potential for patients with cancer undergoing radiation therapy.


Avasopasem manganese (AVA or GC4419), a selective superoxide dismutase mimetic, is in a phase 3 clinical trial (NCT03689712) as a mitigator of radiation-induced mucositis in head and neck cancer based on its superoxide scavenging activity. We tested whether AVA synergized with radiation via the generation of hydrogen peroxide, the product of superoxide dismutation, to target tumor cells in preclinical xenograft models of non–small cell lung cancer (NSCLC), head and neck squamous cell carcinoma, and pancreatic ductal adenocarcinoma. Treatment synergy with AVA and high dose per fraction radiation occurred when mice were given AVA once before tumor irradiation and further increased when AVA was given before and for 4 days after radiation, supporting a role for oxidative metabolism. This synergy was abrogated by conditional overexpression of catalase in the tumors. In addition, in vitro NSCLC and mammary adenocarcinoma models showed that AVA increased intracellular hydrogen peroxide concentrations and buthionine sulfoximine– and auranofin-induced inhibition of glutathione- and thioredoxin-dependent hydrogen peroxide metabolism selectively enhanced AVA-induced killing of cancer cells compared to normal cells. Gene expression in irradiated tumors treated with AVA suggested that increased inflammatory, TNFα, and apoptosis signaling also contributed to treatment synergy. These results support the hypothesis that AVA, although reducing radiotherapy damage to normal tissues, acts synergistically only with high dose per fraction radiation regimens analogous to stereotactic ablative body radiotherapy against tumors by a hydrogen peroxide–dependent mechanism. This tumoricidal synergy is now being tested in a phase I-II clinical trial in humans (NCT03340974).


Superoxide dismutases (SODs) were first described in 1969 as metalloproteins that catalyze the dismutation of two superoxide molecules (O2•−) to hydrogen peroxide (H2O2) and oxygen (O2) (1). SOD (or dismutase) mimetics are small-molecule (~500 g/mol) agents that mimic the activity of endogenous SODs. MnSOD (aka SOD2) is a mitochondrial enzyme containing catalytic manganese and the inspiration for two prominent classes of dismutase mimetics: the Mn-pentaazamacrocyclic and Mn-porphyrins, examples of which are in clinical trials as normal tissue radioprotectors (NCT02655601 and NCT030608020) (2, 3). Avasopasem manganese (AVA), the lead Mn-pentaazamacrocyclic dismutase mimetic, was shown to substantially reduce the incidence, duration, and severity of severe oral mucositis in patients with head and neck cancer treated with radiation and cisplatin in a recent phase 2b randomized trial (NCT02508389) (4).

Less explored are the potential anticancer effects of selective dismutase mimetics. Some evidence supports the hypothesis that Mn-porphyrin dismutase mimetics exhibit radiosensitizing effects (5, 6). Furthermore, preclinical data show that MnSOD enzyme overexpression has antitumor effects that are abrogated by overexpression of H2O2-metabolizing enzymes including catalase and glutathione peroxidases (516). Last, recent work has provided a mathematical model describing a possible mechanism by which MnSOD overexpression could selectively enhance H2O2 flux in cancer cell mitochondria when acting as a tumor suppressor (17).

Stereotactic ablative radiotherapy (SAbR), also referred to as stereotactic body radiotherapy, is rapidly gaining favor due to technical innovations from imaging to dose delivery that limits doses to the normal tissue but allows ablative doses of radiation [>7.5 grays (Gy) per fraction] delivered in a hypofractionated series of five or fewer fractions to tumors. Although initially used in nonresectable lung cancers, the use of SAbR has expanded to replace surgery in a number of disease sites because of its notable clinical results (1821). Ionizing radiation (IR) exposure results in three waves of oxidant generation that contribute to biological responses governing therapy outcomes (22). The initial wave results immediately from the radiolysis of water to form hydroxyl radicals (OH), carbon-centered radicals, O2•−, H2O2, organic hydroperoxides, and other reactive species. A second wave beginning shortly after radiation results from the up-regulation of the reduced form of nicotinamide adenine dinucleotide phosphate oxidase activity, generating O2•− 1 to 24 hours after IR exposure (23, 24). The third wave of oxidant generation appears to involve both inflammatory responses and mitochondrial electron transport chain processes leading to increased O2•− beginning in the days after exposure (22, 25). The production of metabolically produced O2•− in the oxidant waves produced after radiation is believed to be proportional to IR doses within the therapeutic range (26). Although some radiosensitizers attempt to exploit free radical chemistry at the time of radiation, no studies have determined whether O2•− generated after radiation could be exploited to improve cancer therapy responses after SAbR.

O2•− is responsible for a substantial portion of radiation therapy–induced damage to normal cells and tissue while not being as toxic to cancer cells and tumors (2730). For example, by removing O2•−, a selective dismutase mimetic, such as AVA, decreases radiation damage to normal mucosa but does not reduce efficacy against tumors, presumably because AVA does not target OH (4, 31, 32). Conversely, intracellular H2O2 has been suggested to be more toxic to cancer cells than nonmalignant cells (27, 30).

On the basis of this prior knowledge, we hypothesized that AVA would specifically amplify the radiation response of tumor cells by enhancing the generation of H2O2 after high dose per fraction IR consistent with SAbR regimens. We reasoned that because O2•− production is proportional to IR dose but the radical itself is extremely short-lived, the larger daily fraction doses used by SAbR protocols should maximize O2•− and the resulting AVA-catalyzed generation of H2O2, resulting in enhanced tumor cell killing. Clinical trials are now underway to test the efficacy of AVA in combination with SAbR.


AVA does not protect human non–small cell lung cancer cell lines or tumors from radiation

The chemical structure of AVA is depicted in Fig. 1A. Given the reduction of acute radiation–induced mucositis in the hamster cheek pouch (33) and mouse total body irradiation models (34) using an enantiomer of AVA (GC4403), it was important to determine whether AVA might reduce tumor radiation response. First, we conducted in vitro clonogenic survival assays using three non–small cell lung cancer (NSCLC) cell lines. At suprapharmacological concentrations of AVA, NSCLC cells were not protected from radiation exposure (Fig. 1, B to D). This observation was consistent across all three cell lines regardless of their Kirsten rat sarcoma (KRAS), p53, or epidermal growth factor receptor (EGFR) mutational status but is not consistent with previous findings that demonstrated statistically significant radiosensitization of NSCLC tumor cells with AVA (31). (Individual data used to generate the composite data seen here can be found in data file S1.) Second, we conducted tumor growth delay (TGD) experiments where a single dose of AVA was administered 30 min before the irradiation of H1299 cells grown as tumors in the legs of athymic nu/nu mice with a dose of 18 Gy. Tumor response was enhanced; however, a synergistic response to the combination of AVA and radiation was seen (Fig. 2). In a second experiment where a dose of AVA was delivered 30 min before irradiation and an additional dose was delivered on each of the four consecutive days after radiation (five daily doses of AVA total), there was a synergistic enhancement of the radiation response (P = 0.016) (Fig. 2). Not only did this dosing schedule further enhance the response of tumors to radiation, but the result was also complete tumor cures, suggesting that dismutation of O2•− generated in the days after IR also contributed to the synergy. We also showed enhanced radiation responses using this schedule in two additional human NSCLC tumor xenografts, A549 (P < 0.0001 and P < 0.0001 for one and five AVA doses, respectively) and HCC827 (P = 0.0004 and P < 0.0001 for one and five AVA doses, respectively), although the responses were not synergistic (Fig. 2).

Fig. 1 Pretreatment of NSCLC cell lines with AVA does not protect against IR-induced cancer cell killing.

(A) Chemical structure of AVA. (B to D) NSCLC tumor cell lines (B) H1299, (C) A549, and (D) H460 were treated with 24 or 48 μM concentrations of AVA 30 min before irradiation with 0 to 8 Gy of γ-rays. Data shown as means ± SEM of three replicates in four independent experiments.

Fig. 2 AVA enhances TGD in H1299, A549, and HCC827 ectopic xenografts after IR.

Averaged and individual mouse xenograft tumor volumes comparing the effects of AVA alone or combined with a single dose of 18 Gy. H1299 (top two rows), A549 (middle two rows), and HCC827 (bottom two rows) were all evaluated with a single dose of AVA 30 to 60 min before irradiation (left column) and AVA delivered once per day on an additional 4 days after irradiation (right column). All animal cohorts contained n = 8 to 10 animals per group.

The enhanced radiation response with AVA is more pronounced at higher dose per fraction

The dose of 18 Gy was chosen for H1299 TGD experiments to test for any potential for tumor radioprotection or radiosensitization. The notable antitumor effects of the combination seen in Fig. 2 in contrast to the single-agent activity of AVA in these lines suggested additional mechanisms of radiation-induced tumor cell–killing synergy in vivo that relied on high dose per fraction exposures and redox metabolism after exposure. To test this notion, we selected biologically effective dose fractionation of the 18 Gy × 1 fraction, consisting of 9.9 Gy × 2 fractions, 7.3 Gy × 3 fractions, 5 Gy × 5 fractions, and 2 Gy × 16 fractions delivered on consecutive days (Fig. 3). Per design, these fractionation schemes yielded equivalent TGD curves for radiation alone. AVA was effective at enhancing the radiation response only when used with higher fraction doses, with a threshold effect beginning between 7.3 and 9.9 Gy. Synergy was again observed with AVA for the 18 Gy × 1 fraction (P = 7.2 × 10−9) and 9.9 Gy × 2 fractions (P = 0.0036) schedules but not for the 7.3 Gy × 3 fractions, the 5 Gy × 5 fractions, or the 2 Gy × 16 fractions schedules. There was no significant enhancement of tumor response at 5 Gy × 5 fractions nor was there evidence of radioprotection of tumors when treated with 16 fractions of 2 Gy.

Fig. 3 AVA enhancement of radiation-induced TGD increases with increasing radiation dose per fraction.

H1299 xenografts were treated with IR schedules of 18 Gy × 1 fraction, 9.9 Gy × 2 fractions, 7.3 Gy × 3 fractions, 5 Gy × 5 fractions, or 2 Gy × 16 fractions alone (biologically equivalent schedules) and in combination with AVA given once 30 to 60 min before irradiation and once per day for the next 4 days. Average tumor volumes (left column), individual tumor volumes (middle column), and Kaplan-Meier analysis using the IACUC threshold of 1000-mm3 tumor volume as a proxy for survival are shown. All animal cohorts contained n = 8 to 10 animals per group.

The radiation dose necessary for tumor cure is reduced with the addition of AVA

TGD is not a true measure of tumor cure, because any single surviving clonogen can effectively repopulate the tumor, resulting in locoregional failure, in addition to the potential for acquired therapeutic resistance. Therefore, to determine the potential for AVA to enhance the tumor cure rate, we also performed tumor cure dose to achieve 50% tumor kill [50% tumor control dose (TCD50)] assays in H1299 xenografts using radiation alone or in combination with AVA at 10 or 24 mg/kg before and for each of the next 4 days after irradiation (Fig. 4). On the basis of logistic fit regression analyses, the TCD50 radiation dose shifted from 24.6 Gy for radiation alone to 14.7 and 19.6 Gy with dose enhancement factors of 1.62 and 1.22, for combinations with 24 and 10 mg/kg of dose of AVA, respectively.

Fig. 4 AVA decreases the TCD50 dose for H1299 xenografts exposed to IR.

H1299 tumor xenografts were treated with single doses of radiation alone or with either 10 or 24 mg/kg of AVA given 30 to 60 min before irradiation and once daily for the next 4 days. Tumor growth was followed for 120 days after radiation exposure, and tumor cure was determined by necropsy. The horizontal bars reflect the 95% confidence interval for each curve at a relative tumor cure value of 0.50 (TCD50). All animal cohorts contained n = 10 to 12 animals per group.

AVA increases steady-state concentrations of H2O2 in cancer cells, and overexpression of catalase in H1299 xenografts diminishes the enhanced radiation response seen with AVA

Given the rapid rate of AVA-catalyzed conversion of O2•− to H2O2 and previous studies with native MnSOD enzyme overexpression (17) and exogenous recombinant MnSOD (28) that demonstrated elevated concentrations of H2O2, we hypothesized that increased concentrations of H2O2 in tumor cells contributed to the synergistic therapy responses seen with SAbR doses of radiation combined with AVA. We first demonstrated in vitro with H1299 lung cancer cells that AVA increased steady-state concentrations of intracellular H2O2 (Fig. 5A). Furthermore, inhibiting H2O2 metabolism during exposure to AVA by simultaneously depleting glutathione [with buthionine sulfoximine (BSO)] and inhibiting thioredoxin reductase activity [with auranofin (Au)] selectively sensitized H1299 cells to AVA-mediated killing. The combination, however, did not increase normal human bronchial epithelial cell (HBEpC) killing (Fig. 5B), supporting the selective toxicity of intracellular H2O2 generated by AVA in cancer cells versus normal cells. Furthermore, the enhanced sensitivity to AVA combined with BSO/Au was inhibited by the thiol antioxidant N-acetylcysteine (NAC) demonstrating that thiol oxidation contributed to these biological effects (Fig. 5C). To determine the generality of these biological effects on H1299 lung cancer cells, we repeated these experiments using MB231 triple-negative breast cancer cells. As was seen with H1299 cells, AVA increased steady-state concentrations of H2O2 in MB231 cells (fig. S1A) and inhibition of H2O2 metabolism (by BSO and Au) selectively sensitizing MB231 cells to AVA killing relative to normal human mammary epithelial cells (HMECs), a sensitization that was inhibited by NAC (fig. S1, B and C).

Fig. 5 AVA increases the steady-state concentration of H2O2 leaving cells susceptible to modulation of redox metabolism in vitro AVA.

(A) H1299 cells were grown exponentially in vitro for 2 days and then assayed for intracellular steady-state concentrations of H2O2 for 60 min in the presence of 20 μM AVA. H2O2 concentration (picomolars per cell) was normalized to the untreated control in each experiment (n = 3, *P = 0.006, one-tailed t test). (B) HBEpC and H1299 cells were grown exponentially for 2 days and then treated with BSO (100 μM) and AVA (20 μM) for 48 hours with Au (500 nM) added for the last 15 min immediately before the clonogenic assay (N = 3, **P < 0.0001 compared to HBEpC). (C) H1299 cells were grown exponentially 2 days then treated with AVA (20 μM), BSO (100 μM), or NAC (10 mM) for 48 hours with Au (500 nM) added for the last 15 min before clonogenic assay (α = P < 0.0001, one-tailed t test). (D) The doxycycline induction of catalase activity in H1299-CAT tumors harvested after 3 days of doxycycline treatment (2.5 mg/ml in drinking water). (E) Growth delay of H1299-CAT xenografts exposed to AVA alone, a single 18-Gy IR dose, or 18-Gy IR plus AVA before IR and for 4 days after IR, without (left column) or with doxycycline (right column) in the water (n = 8 animals per group).

Given these in vitro studies, stable lentiviral transduction into H1299 cells of a doxycycline-inducible construct for conditional overexpression of catalase was developed (H1299-CAT). Xenografts grown from H1299-CAT cells demonstrated a 5- to 10-fold increase in catalase activity when doxycycline was added to the drinking water (Fig. 5D). TGD experiments demonstrated that without doxycycline in the drinking water, H1299-CAT tumors had similar growth and response to high dose per fraction IR consistent with SAbR (18 Gy × 1) compared to parental H1299 line tumors; AVA administered before and for the next 4 days also enhanced the effects of IR (Fig. 5E). However, when doxycycline was added to the drinking water and catalase was then overexpressed, the synergistic effect of the combination of AVA and high dose per fraction IR was almost completely eliminated (Fig. 5E). Because catalase is an enzyme that specifically scavenges H2O2, these results provide strong evidence supporting that the interaction between AVA and SAbR-like radiation therapy is mediated by H2O2. [The individual data used to generate the composite data seen in Fig. 5 (A to D) can be found in data file S2.]

The enhanced response of tumors to radiation with AVA is not specific to NSCLC

Many tumor types may have decreased concentrations of SOD expression and activity (30, 35, 36), decreased concentrations of catalase expression and activity (30), translocation of catalase (27), or other susceptibility to elevated intracellular H2O2 (37), and as seen in vitro, AVA was able to increase steady-state concentrations of H2O2 in both human lung and breast cancer cells. To test the generality of the synergy between AVA and SAbR in xenografts from several human cancer types, we selected the radioresistant SqCC/Y1 head and neck squamous cell carcinoma (HNSCC) cell line because AVA has demonstrated oral mucosa radioprotective efficacy in patients receiving treatment for HNSCC, including in a phase 2B clinical trial (4). In addition, we chose the pancreatic ductal adenocarcinoma (PDAC) cell lines Panc 02.03, SW1990, and PANC-1 due to an ongoing phase 1b/2a clinical trial examining the effects of AVA in combination with SAbR in patients receiving treatment for pancreatic cancer (NCT03340974). As was seen with NSCLC xenografts, high dose per fraction IR (12 Gy × 1) with AVA enhanced the response of tumors in SqCC/Y1 (P = 0.0184), Panc 02.03 (P = 0.0023), SW1990 (P = 0.0376), and PANC-1 (P = 0.035) tumors (Fig. 6). The response of Panc 02.03 tumors was synergistic (P = 0.041) when treated with AVA and radiation.

Fig. 6 AVA enhancement of tumor response to ablative radiation doses is not specific to NSCLC xenografts.

Growth delay of xenografts exposed to a single 12-Gy IR fraction with or without AVA using human cancer cell lines SqCC/Y1 (HNSCC), Panc 02.03, SW1990, and PANC-1 (all PDAC).

Transcriptome profiling of H1299 xenografts by treatment group

Total RNA sequencing was performed on H1299 tumors treated either with or without a single dose of 18 Gy fraction and with or without AVA given 30 min before radiation and daily for additional 4 days. RNA was extracted from three tumors from each treatment group at 1 hour after the initiation of treatment and subsequently on days 1, 3, and 7. Unsupervised gene set enrichment analysis (GSEA) identified seven hallmark pathways that were altered in irradiated tumors treated with AVA (Figs. 7 and 8 and figs. S3 and S4) compared to irradiated tumors. Boxplots of mean activation z scores for each pathway and heatmaps of log expression values for each gene in the pathway over time describe the temporal changes in hallmark pathways for irradiated tumors treated with AVA compared to irradiated only tumors. Substantial differences were seen between the cohorts at hour 1 and day 1, but by day 3, the differences were reduced, and by day 7, activation z scores were no different, potentially reflecting AVA unavailability because the last dose of AVA was delivered on day 4. Whereas activation z scores for hedgehog signaling (Fig. 7A) suggested the early down-regulation of this pathway, activation z scores suggested that apoptosis (Fig. 7B), myogenesis (fig. S3A), inflammatory response (Fig. 8A), epithelial to mesenchymal transition (EMT) signaling (fig. S3B), hypoxia (fig. S4), and tumor necrosis factor–α (TNFα) signaling via nuclear factor κB (NFκB) (Fig. 8B) were up-regulated early on in the IR-plus-AVA cohort compared to the IR-only cohort. Leading edge analysis enrichment scores examined at day 1 across all pathways confirmed the direction of the activation z scores at day 1 as seen in Figs. 7 and 8 and figs. S3 and S4.

Fig. 7 Hedgehog and apoptosis signaling pathways are differentially regulated after irradiation of H1299 xenograft tumors depending on AVA treatment.

Tumors were treated with a single fraction of 18-Gy IR alone or 18 Gy + AVA. (A) Hedgehog signaling and (B) Apoptosis signaling, are described by activation z scores, heatmaps, and leading edge analysis, from the top to bottom of each panel. Boxplots of activation z scores for each treatment cohort (median, 75th, and 99th percentiles) are plotted over their respective column of the heatmap of log gene expression for each gene associated with the hallmark pathway. The results of leading edge analysis for samples at day 1 after irradiation confirmed the enrichment direction for each pathway including the nominal P value and FDR at that time.

Fig. 8 GSEA identified inflammatory and TNFα/NFκB pathways as differentially regulated in irradiated H1299 tumors treated with AVA over time after irradiation.

(A) Inflammatory signaling, and (B) TNFa via NFKb signaling, are described by activation z scores, heatmaps and leading edge analysis, from the top to bottom of each panel. Boxplots of activation z scores for each treatment cohort (median, 75th and 99th percentiles) are plotted over their respective column of the heatmap of log gene expression for each gene associated with the hallmark pathway. The results of leading edge analysis for samples at day 1 after irradiation confirm the enrichment direction for each pathway including the nominal P value and FDR at that time. Tumors were treated with a single fraction of 18-Gy IR alone or 18 Gy + AVA.

Supervised expression analysis identified 388 genes, including 62 long noncoding RNAs, that exhibited different expression profiles based on treatment group and time after irradiation. Canonical pathway analysis via the Ingenuity Pathways Analysis (IPA) platform using the 388 genes from the supervised expression analysis identified altered canonical pathways that aligned well with the GSEA hallmark pathway analysis including cell survival or cell death signaling, cytokine signaling, and inflammatory and immune signaling as examples (fig. S2).


This is a preclinical study testing the hypothesis that selective SOD mimetics, which have been shown clinically to reduce normal tissue radiation toxicity (4, 32), can also be used in combination with high dose per fraction radiotherapy regimens such as SAbR to enhance tumor killing. We demonstrated that these dual radiation response modifiers act through a mechanism that exploits fundamental differences in O2•− and H2O2 metabolism between cancer and normal cells. The selective dismutase mimetic AVA exhibits specificity for O2•−, and by removing it has clinically demonstrated reduction in normal tissue toxicity. At the same time, in converting O2•− to H2O2, this mechanism allows for a robust radiosensitizing effect for dismutase mimetics in cancer cells. The current study clearly shows that AVA enhanced the response of H1299 tumors when radiation is delivered at doses above an approximately 7 Gy per fraction threshold. In addition, this enhancement increased with increasing radiation dose and limited fraction number because the dose and fractionation schedules represented biologically equivalent TGD responses. Furthermore, we also showed that dosing AVA for several days after irradiation amplified the antitumor effect, suggesting that substantial O2•− was still being generated.

When AVA was combined with the highest dose of radiation, the response was synergistic and was dependent on the generation of H2O2, because catalase overexpression substantially abrogated the radiosensitizing effect of AVA. Synergy was observed among all of the H1299, NSCLC, and Panc 02.03 pancreatic tumors examined, suggesting that the potential for a synergistic response to the combination of radiation and AVA is not tissue specific. Furthermore, there may be a dose threshold effect for synergy given that synergy was identified for H1299 tumors treated with either 18 or 9.9 Gy in each of two fractions. Synergy was not seen in H1299 tumors irradiated with fraction sizes smaller than 9.9 Gy. Moreover, a TCD50 assay determined that there was an AVA dose enhancement factor of 1.62, which is generally more than that seen with other agents used in combination with radiation, such as gemcitabine (1.4 to 1.54) (37, 38), C225 (1.8) (39), cisplatin (1.3 to 1.4) (40), and nimorazole (1.3) (41), which are limited in their use by increases in treatment toxicity.

Given the reduction in oral mucositis seen with selective dismutase mimetics in the hamster cheek pouch model and in the human head and neck cancer clinical trials (4, 33), as well as the antitumor response seen here with high dose per fraction IR, there exists evidence for a dual functionality of Mn-pentaazamacrocyclic dismutase mimetics to both reduce radiation-induced normal tissue toxicity by removing O2•− and increase antitumor SAbR responses. We propose that the differential cell-killing responses between tumor and normal cells could be a combination of any of at least three mechanisms. First, as previously proposed, there are the differences in endogenous MnSOD activity seen in normal cells versus tumor cells (30). According to this model, H2O2 flux is reduced in cells with low endogenous SOD activity, and decreased MnSOD concentrations have been observed in many primary cancers. Others have also reported that SOD proteins may be present at normal concentrations but with loss of activity, for example, due to posttranslational modification (35, 36). Thus, it could be expected that enhancement of dismutase activity with AVA would have a greater relative effect on H2O2 flux in tumor cells than normal cells. Second, tumor cells exhibit fundamental differences in mitochondrial metabolism leading to higher steady-state concentrations of O2•−. It has also been suggested that higher steady-state concentrations of O2•− in tumor cells relative to normal cells occur as a result of alterations in the stoichiometry of the assembly of electron transport chains (42), allowing AVA in cancer cells to increase the flux of electrons to form H2O2 via Le Chatelier’s principle (17). In addition, transient increased reactive oxygen species (ROS) generation via collapse of mitochondrial membrane potential can trigger ROS-induced ROS release (RIRR), releasing ROS into the cytosol, which induces RIRR in neighboring mitochondria, leading to a cascade of enhanced ROS production and the potential for further mitochondrial or cellular injury (43). Last, impaired intracellular H2O2-detoxifying capacity combined with higher concentrations of labile iron pools in tumor cells as a result of increased steady-state concentrations of O2•− may enhance the rate of OH generation relative to normal cells treated with AVA (44, 45). Whatever the mechanism behind tumor H2O2 sensitivity, it is tempting to speculate that the approximately 7 Gy per fraction threshold, where AVA antitumor synergy with radiation begins to substantially increase, reflects a point where additional H2O2 overwhelms cancer cells’ tolerance for the metabolic production of oxidants.

With respect to alterations in molecular signaling at the level of tumor tissue, gene expression analysis revealed changes in redox-associated signaling and cell survival/death pathways, although sample numbers and time after irradiation were limited. Changes in redox-associated signaling were found in the analysis of temporal changes in gene expression in the presence of AVA whether tumor cells were irradiated. For example, in AVA only–treated tumors, NFκB signaling was up-regulated compared to controls. Interleukin-6 signaling, which is driven by NFκB signaling, was also up-regulated in AVA-treated tumors compared to the other treatment cohorts. Whether this is linked to H2O2 production and signaling (46) is unknown. However, as limited as the data are, the results did identify signaling changes that one might not have considered, particularly in the context of radiation therapy. For example, the top canonical pathways identified using the IPA platform included an across-the-board temporal reduction of cholesterol biosynthesis in tumors treated with radiation only or with radiation plus AVA. Cholesterol plays important roles in tumor cell growth, metabolism, and intracellular signaling including activation of the hedgehog pathway (47). Cholesterol metabolism is modulated by the Liver X Receptor/Retinoid X Receptor (LXR/RXR) pathway (48) but is uncoupled from LXR/RXR regulation in cancer cells via transcriptional down-regulation of LXR/RXR. Although tumors exposed to AVA appeared to up-regulate LXR/RXR activation signaling and stimulate cholesterol metabolism, radiation exposure down-regulated cholesterol metabolism irrespective of AVA exposure or LXR/RXR activation. Regarding the former, the activation z score of the GSEA hallmark pathway for hedgehog signaling, abnormally activated in cancer cells stimulating tumor cell proliferation and multidrug resistance (49), was lower in irradiated tumors treated with AVA compared to irradiated-only tumors. Leading edge analysis at 1 day after irradiation supported down-regulation of that pathway. Also seen was the apparent increased activation z score at day 3 and leading edge enhancement score, both of which suggested an up-regulation of the myogenic pathway. This is an intriguing finding, particularly given that O2•− and H2O2 play important roles in myogenesis including myogenic differentiation and arteriole contraction processes (50). Although the processes leading to muscle atrophy in patients with cancer are complex, one could speculate that AVA might have a positive effect by up-regulating myogenic processes.

The early up-regulation of the inflammatory response, apoptosis signaling, and the down-regulation of NFκB signaling suggest that adding AVA to IR produces a more rapid and potent tumor response via reduced survival pathway signaling and enhanced cell death pathway signaling. A similar perturbation in survival/cell death pathways has also been reported for lung cancer cells (31). This is consistent with the tumor model results here and with respect to H2O2 driving the synergy between AVA and SAbR-like radiation but stands in contrast with the published anti-inflammatory properties of Mn-pentaazamacrocyclic dismutase mimetics, such as AVA in normal tissue (51), further emphasizing the differential effects of AVA in cancer and normal cells. Last, it is worth pointing out that the in vivo studies described here were carried out in nude, that is, immunocompromised mice. It is conceivable that in an immunocompetent setting, there may be additional immunological effects that were not captured here because both O2•− and H2O2 can function as second messengers in immune cell signaling (52).

In summary, that the extent to which the enhanced nonclinical antitumor effect described herein with a Mn-pentaazamacrocyclic dismutase mimetic will translate clinically to more effective SAbR regimens or to decreased IR dose thresholds that are sparing of normal tissue will only be borne out by human clinical trials. However, the phenomena and method of action described here are supportive of such clinical trials. The data show that the antitumor synergy of the combination may also apply to sites other than NSCLC given the response of pancreatic tumor xenografts6 (Fig. 6). The PDAC clinical trial NCT03340974 is based on the antitumor effect seen here. In addition, examination of the role that redox regulation by selective dismutase mimetics may play in anticancer immune responses in irradiated tumors could point to additional potential therapeutic application. Previous studies combining dismutase mimetics with radiation (53) and the enrichment of inflammatory gene signatures seen here suggests a potential additional immunological component to the synergistic response of preclinical tumor models treated with SAbR-like radiation plus AVA.

Last, the predominant limitations of the current studies are twofold. First, we acknowledge that the in vivo experiments used tumor xenografts of human tumor cell lines grown and treated in nude mice. Thus, there are questions of clinical translation of the results; however, the results of ongoing clinical trials for pancreatic cancer (NCT03340974) and lung cancer (NCT04476797) will address these concerns. Second, because these studies were conducted in athymic nu/nu mice that lack functional T cells, the results presented in the absence of a complete immune system may not reflect outcomes in humans. Additional preclinical studies are ongoing to examine the potential for AVA enhancement of radiation response where the preclinical models include functional immune systems.


Study design

Initial efforts to ensure that AVA (a radioprotector) was not protective of tumors led to the utilization of human NSCLC, HNSCC, and PDAC tumor lines in clonogenic survival assays and then as human tumor xenografts in athymic nu/nu mice. Once the efficacy of AVA to enhance the radiation response of tumors to high dose per fraction irradiation was established, studies including those measuring hydrogen peroxide and transcriptomics of tumor xenografts were performed to establish the mechanism of action. Individual tumor volume was tracked on a per animal basis using physical measurement, and ex vivo tumor samples were collected for hydrogen peroxide quantification, catalase activity detection, and transcriptomic analyses. All group sizes for animal experiments were determined via power analysis with α values set to 0.05 and β values set to 0.8. These calculations determined that eight animals per group were more than sufficient to detect significant differences with appropriate power. For all experiments, animals were double-blind randomized among treatment groups once tumor volumes reached a 100- to 200-mm3 volume. Measurements were taken by a blinded individual and unblinded by a second individual.

TGD and TCD50 studies

Animals were purchased from Charles River and maintained and handled under protocols approved by the Institutional Animal Care and Use Committee (IACUC) at the University of Texas Southwestern Medical Center. For experiments using human tumor xenograft lines, 6-week-old female athymic nu/nu mice were inoculated with 5 × 106 human tumor cells (H1299, A549, HCC827, SqCC/Y1, Panc 02.03, SW1990, PANC-1, and SqCC/Y1) suspended in serum-free culture medium subcutaneously to the right hip. Tumor volume was measured using calipers two to three times weekly. Tumor treatment was initiated when tumor volume ranged between 100 to 150 mm3, and volume was tracked until euthanasia (tumor size exceeding no more than 2 cm in any one dimension) or 120 days after treatment, at which point, if no mass was present, the tumor was considered “cured.”

Mathematical determination of synergy

Synergy was determined on the basis of the Bliss definition of drug independence as previously described (54). Briefly, individual TGD curves for each animal in a treatment cohort were log-transformed and fitted using a linear regression. The slope values were then put into the equation [Radiation] + [AVA] – [Radiation and AVA combination] – [Control]. If the final value was greater than 1.0, then the combined response was considered synergistic.


Irradiations for in vitro cell culture experiments were performed using a Mark 1 sealed 137Cs source irradiator (J.L. Shepherd and Associates). Animal irradiations were performed using an X-Rad 320 irradiator (Precision X-Ray) running at 250 kilovolt peaks and 15 mA. Tumor irradiations were performed while animals were under anesthesia, and radiation was targeted to focally irradiate the tumor using a 10-mm collimated beam to spare at-risk normal structures. For studies using fractionated exposures, fractions were delivered on consecutive days.

Drug handling and in vivo delivery

AVA was provided by Galera Therapeutics. AVA was solubilized to a concentration of 24 mM in bicarbonate buffered saline solution at pH 7.4 and delivered as a single injection intraperitoneally as 24 mg/kg of dose 30 to 60 min before irradiation. The human equivalent dose would be about 90 mg, which is in the range of dose received by patients (90 to 100 mg/day) in the clinical trials described above. Doxycycline hydrochloride (VWR) was solubilized to a concentration of 2.5 mg/ml in distilled deionized water with 1% sucrose.

Cell lines and cell culture

Cell lines H1299, A549, H460, MB231, HCC827, PANC-1, Panc 02.03, and SW1990 were purchased from the American Type Cell Culture repository. The SqCC/Y1 cell line was obtained from MD Anderson Cancer Center. Human lung adenocarcinoma cell lines H1299, A549, H460, and HCC827 were cultured in Dulbecco’s modified Eagle’s medium (DMEM)/F12 basal medium (Sigma-Aldrich) supplemented with 10% fetal plus fetal bovine serum (FBS) (Atlas Biologicals) and l-glutamine (Sigma-Aldrich). Human PDAC lines PANC-1 and SW1990 and HNSCC line SqCC/Y1 were cultured in DMEM with high glucose (Sigma-Aldrich) supplemented with 10% FBS and l-glutamine. PDAC cell line Panc 02.03 was cultured in RPMI 1640 basal medium (Sigma-Aldrich) supplemented with 15% FBS, l-glutamine, and human recombinant insulin (10 U/ml; Gibco). All cell culture was conducted in a class two, certified biological safety cabinet, and cells were grown at 37°C and 5% CO2 in a humidified incubator. Cell cultures were routinely tested for mycoplasma and were genotyped to confirm their identity.

Radiation clonogenic survival assays in vitro

Cells were plated into 60-mm2 dishes and allowed to adhere for 6 hours. After attachment, cells were irradiated with γ-ray doses of 2, 4, 6, and 8 Gy. Cells were returned to incubation and allowed to divide for 10 population doublings (7 to 9 days depending on cell line). Plates were then stained with a solution containing 0.5% crystal violet dissolved in 20% methanol. Colonies were counted using a dissecting microscope, and surviving fraction was determined on the basis of the number of colonies containing 50 or more cells. Results indicate the compilation of at least three individual experiments.

Tumor versus normal cell clonogenic survival assays with AVA

A total of 1 × 105 H1299 cells, 2.5 × 105 nontransformed HBEpCs (Cell Applications Inc.), 1.25 × 105 MB231 triple-negative human breast carcinoma cells (a gift from M. Henry, University of Iowa), or 3.0 × 105 nonimmortalized HMECs (Clonetics) were plated on 60-mm dishes for 72 hours at 4% O2, 5% CO2, and 37°C. All cells were then washed with warm phosphate-buffered saline (PBS) and placed in either complete bronchial/tracheal epithelial cell basal medium (H1299 cells and HBEpCs) or mammary epithelial cell basal medium (HMECs and MB231 cells) (Cell Applications Inc.). Cells were then treated with AVA, Au (Enzo Life Sciences), and BSO (Sigma-Aldrich) for 24 hours, followed by clonogenic survival assay as described (55). HBEpC cloning dishes also contained 1 × 105 lethally irradiated (30 Gy) H1299 feeder cells, which were also separately plated without HBEpCs to confirm the lack of colony formation. After treatment, attached cells were trypsinized with 0.25% trypsin-EDTA, inactivated with media, containing 10% FBS, counted using a Beckman Coulter Counter, and plated at densities ranging from 200 to 2000 cells per dish. Clones were grown for 10 to 12 days in complete medium with 0.1% gentamycin. Cells were fixed with 70% ethanol and stained with Coomassie blue, and colonies containing >50 cells were counted. The treatment groups for each cell line were normalized to the control group. The survival analysis was performed using a minimum of three cloning dishes per experimental condition, and the experiments were repeated on a minimum of three separate occasions.

Hydrogen peroxide quantification

To quantify intracellular H2O2 during exposure to AVA, a method based on the stoichiometric inactivation of catalase by 3-amino-1,2,4-triazole (3AT) in the presence of H2O2 was performed as previously described (56). The pseudo first-order rate constant for the 3AT-mediated inactivation of catalase was then derived through kinetic analysis. Because endogenous catalase activity in H1299 and MB231 cells is relatively low, the cells were first transduced with 50 multiplicity of infection replication-incompetent adenoviral vector AdCMV-Catalase (ViraQuest) in serum-free medium for 6 hours at 37°C at 21% O2. The medium was then changed to 10% FBS, and cells were recovered at 37°C for 48 hours in the absence of virus. After 48 hours, the cells were rinsed in PBS, and FBS-free medium was added. Fifty millimolars of 3AT was added to the cultures for 5 min before 20 μM AVA. A positive control was generated by treating cells with 100 μM H2O2. Cells were harvested by rinsing twice in ice-cold PBS, scraping into catalase buffer [50 mM potassium phosphate (pH 7.0)], and quickly frozen until analysis of catalase activity. Control and AVA-treated cells were harvested at 0, 10, 20, 30, 50, and 60 min, whereas the H2O2-treated dishes were harvested at 0, 5, 10, and 15 min. Two dishes were assayed at each time point, and the experiment was repeated on three separate occasions.

Generation of H1299-CAT

H1299-CAT cells were generated similarly to as previously described (44, 57). To generate H1299-CAT cells, lentivirus was produced in the TSA201 cell line using pCMV-VSV-G and psPAX2 helper vectors (Addgene). H1299 cells were plated and allowed to grow for 48 hours, and then virus was added to cells with polybrene (8 μg/ml) every 24 hours for 2 days. After transduction, cells were selected with puromycin (10 μg/ml). Surviving cells were plated in 150-mm dishes with 1000 cells per dish. Clones were grown for 10 days, and then several colonies were picked and expanded. To test for maximal catalase overexpression in vitro, cells were treated with doxycycline (1.5 μg/ml) for 48 hours. Protein concentration was determined using the Lowry assay. Increased catalase activity was verified by measuring the decomposition of H2O2 by cell lysates. Maximal activity was found in clone 26, which was used for these experiments.

Collection of tumors for total RNA sequencing analysis and isolation of tumor total RNA

Tumors from three animals in each treatment cohort were collected after 1 hour and on days 1, 3, and 7 after irradiation and/or initiation of AVA treatment. H1299 tumor tissue was collected, 25 mg of tumor was homogenized in QIAzol lysis reagent (QIAGEN), and samples were immediately flash-frozen in liquid nitrogen and stored at –80°C. Once all samples had been collected, the miRNeasy Mini Kit (QIAGEN) was used to isolate total RNA. Once isolated, RNA concentration was determined using a NanoDrop 2000 spectrophotometer (Thermo Fisher Scientific), and quality was assessed using an Experion electrophoresis system (Bio-Rad).

Transcriptome analysis of tumor RNA

Sequencing library was prepared using the Illumina TruSeq Stranded Total RNA Prep Kit. Single-end next-generation sequencing was performed by DNALink using an Illumina NextSeq500 sequencer with an average coverage of 48.6 M. Sequencing reads were trimmed to remove adaptor sequences and aligned with human genome GRCh38 using the Spliced Transcripts Alignment to a Reference (STAR) package with the 2-pass option, which limits mismatches to no more than 10 by default. The aligned reads were quantified using the RNA-Seq by Expectation-Maximization (RSEM) program. Gene counts were summarized and normalized using tximport (58) and DESeq2 (59). Data processing was performed using the Stampede2 supercomputer at the Texas Advanced Computing Center. Genes with very few reads were removed by the default values of the filterByExpr function within the edgeR package (60) that kept genes with a count per million of >0.2 in at least two samples and total read counts of >15 across all samples.

GSEA was performed using GSEA 4.0.1 (61). Genes from the whole transcriptome profiles were ranked on the basis of P values derived from Wald’s test comparing the radiation alone group and the radiation plus AVA group. Top GSEA hallmark pathways were identified by selecting P < 0.05.

Supervised gene lists that differentiated experimental cohorts in this study were obtained by the time series analysis using DESeq2 with a false discovery rate (FDR) of <0.2. Pathway analysis of the supervised gene list was performed using IPA software (QIAGEN). Top canonical pathways were selected by applying an FDR of <0.05 and z score of >1 in Fisher’s exact test. Normalized counts for each gene were centered and converted to z scores to generate heatmaps.

Catalase activity assays

Tumors were homogenized in 200 μl of 10 mM Diethylenetriaminepentaacetic acid (DETAPAC) in phosphate buffered saline (PBS) buffer, sonicated (3 × 5–s on and 10-s off in water bath sonicator), and protein concentration was determined using Pierce BCA Protein Assay Kit (Thermo Fisher Scientific). Catalase activity was determined at 25°C according to the method of Beers and Sizer with the analysis of Aebi (6264). Briefly, 25 μl of samples were added to 4.0 ml of phosphate buffer, mixed, and separated into two quartz cuvettes, blank and active. An H2O2 working solution was added to the active cuvette, and the rate of disappearance of absorption at 240 nm was measured. Bovine catalase was used for the standard. All activity assays were normalized per milligram of cellular protein.

For cell culture experiments, samples were thawed, homogenized in potassium phosphate buffer (pH 7.0), sonicated, and quantified using the Lowry assay. Catalase activity was determined on whole homogenates using the method of Beers and Sizer with the analysis of Aebi (3436) and normalized to protein content.

Statistical analysis

In vitro experiments were carried out at least in duplicate and significance determined by Student’s t test or by one-tailed t test, and P < 0.05 was considered significant. Given the limited number of replicates, the data were not formally tested for normality. The number of biological replicates, number of independent experiments, and statistical tests used can be found in the relevant methodology sections and in the figure captions.


Fig. S1. AVA increases steady-state concentrations of H2O2 and inhibition of glutathione- and thioredoxin-dependent H2O2 metabolism selectively sensitizes MB231 breast cancer cells to AVA in vitro.

Fig. S2. Canonical pathways identified as overrepresented using a 388-gene signature that separates treatment cohorts.

Fig. S3. AVA treatment results in differential regulation of myogenesis and EMT pathways after irradiation of H1299 tumors.

Fig. S4. AVA treatment results in differential regulation of hypoxia pathways after irradiation of H1299 tumors.

Data file S1. Compilation data for Fig. 1.

Data file S2. Compilation data for Fig. 5 (A to D).


Acknowledgments: We would like to thank the Radiation and Free Radical Research Core in the Holden Comprehensive Cancer Center and the Preclinical Radiation Core Facility at UT Southwestern Medical Center. We would like to thank K. Mason for the SqCC/Y1 HNSCC tumor line. Last, we thank E. Demidenko for assistance with statistical tests for synergy. Funding: This work was supported by Galera Therapeutics Inc. and NIH grants T32 CA078586 (to D.R.S.), P30CA086862 (to D.R.S.), P01CA217797 (to D.R.S.), R44CA206795 (to J.L.K. and M.D.S.), and the David A. Pistenmaa M.D., Ph.D. Distinguished Chair (to M.D.S.). Author contributions: M.D.S., B.J.S., and D.R.S. conceived and designed the experiments. B.J.S., T.-K.N., C.D.H., S.N.R., J.D.S., M.A.F., D.S., and C.F.P. performed the experiments. B.J.S., M.D.S., R.A.B., D.P.R., J.L.K., and D.R.S. contributed reagents, materials, and data analysis. B.L. performed statistical analyses to test for synergy. L.D. performed the transcriptome analysis. B.J.S., M.D.S., D.R.S., M.A.F., S.N.R., C.F.P., B.L., C.D.H., and L.D. wrote the manuscript, and all authors edited the manuscript. Competing interests: D.P.R., J.L.K., and R.A.B. are employed by and hold equity interests in Galera Therapeutics Inc., which provided the Mn-pentaazamacrocyclic dismutase mimetic AVA used in this study. D.R.S. and M.D.S. have Sponsored Research Agreements supported by Galera Therapeutics Inc. in preclinical studies of AVA in cancer therapy. D.R.S. is a consultant/advisory board member for Galera Therapeutics. M.A.F., D.R.S., R.A.B., D.P.R., and J.L.K. are inventors on patent application PCT/US2017/030871 submitted by Galera Therapeutics LLC that relates to the combination of pentaazamacrocyclic dismutase mimetics with high dose per fraction radiation and with pharmacologic inhibitors of hydrogen peroxide metabolism for cancer treatment. R.A.B., D.P.R., and J.L.K. are inventors on patent application PCT/US2018/027588 submitted by Galera Therapeutics LLC that relates to the combination of pentaazamacrocyclic dismutase mimetics with high dose per fraction radiation, with immune checkpoint inhibitors, and with both checkpoint inhibitors and high dose per fraction radiation for cancer treatment. All other authors declare that they have no competing interests. Data and materials availability: All data associated with this study are present in the paper or the Supplementary Materials. RNA transcript data and sequencing reads are available at under BioProject accession number PRJNA668807. GC4419 was provided to M.D.S. and D.R.S. under a material transfer agreement with Galera Therapeutics.

Stay Connected to Science Translational Medicine

Navigate This Article